what is stable power circuit in automotive(car)?

Kim-JiHoon

Joined May 3, 2020
63
hello i started my project few month ago.

it's very simple. in automotive(car) FAN control with temperature sensor

my car bettery(24v) -> MCU -> FAN (pwm)

MCU (atmega328p-AU)
FAN (24v / 3A)

MCU need just 50mA current so i use lm7805 instead of lm2576-5v

this is my circuit.

1. POWER (C2, C5 smd type)

2. MOTOR

3. MCU

it works good.

but recently i got problem.

this is problem situation.

when key-on car.

1. MCU is running.

2. temp goes down

3. FAN running (8Khz / 60% pwm duty)

4. car start-up

5. FAN pwm duty goes up 100% (and can't controlling)

6. MCU OFF, still FAN running

7. all shut down, and re start -> it works fine

this problem comes out when FAN is running first and start-up

Q1) what is problem with my power circuit?
(i think the noise when start-up is the problem)

Q2) do i add cap (1000uF/50v) in MOTOR part?
(Image 2.motor -> MOTOR+ - cap - ground)

Attachments

• 260.6 KB Views: 6

wayneh

Joined Sep 9, 2010
17,152
In your drawing, D2 is shown reversed bias. That's not your circuit problem, but it jumps out at me.

C5 should be eliminated but that isn't likely the problem either. If you want to leave it there, you should add a reversed diode around the regulator to discharge the capacitor when Vout > Vin. The datasheet recommends it to prevent damage to the 7805. Most people seem to leave it out and get away with it.

it works good.

but recently i got problem.
Are you saying it worked fine for a while and now is misbehaving? Or did you only notice it recently?

LowQCab

Joined Nov 6, 2012
517
Brushed-DC Motors have a nasty habit of putting out huge amounts of RFI-Hash (noise).
You may be having problems because of this.

D-9 is only rated for 1-Amp, it may already be smoked.
It needs to be rated substantially higher than the MEASURED full-load Amps of the Motor.

A Micro-Controller doesn't have enough Current Output to switch a large FET
with any respectable speed,
although the FET you are using is kinda wimpy.
It's probably running quite hot.

How are your Grounds physically arranged ?
Any Grounds using the Sheet-Metal Body of the Car ?
What kind of Temp-Sensor are you using ?
How is it Grounded ?
Is it a 1-wire, 2-wire, or 3-wire Sensor ?
Does the Temp-Sensor get its Ground from the Engine-Block ?
Are the Wires Shielded ?
Is your Micro-Controller mounted in a Metal-Box ?

Here are some ideas to make you think .................. (picture)

Last edited:

Kim-JiHoon

Joined May 3, 2020
63
In your drawing, D2 is shown reversed bias. That's not your circuit problem, but it jumps out at me.

C5 should be eliminated but that isn't likely the problem either. If you want to leave it there, you should add a reversed diode around the regulator to discharge the capacitor when Vout > Vin. The datasheet recommends it to prevent damage to the 7805. Most people seem to leave it out and get away with it.

Are you saying it worked fine for a while and now is misbehaving? Or did you only notice it recently?
when i reconnected device, it works good!

but i used this about 1~2hour and all shutdown, after 1hour (for a launch) and start-up.

motor goes too high (like maximum over pwm duty setting) and can't control with controller

Brushed-DC Motors have a nasty habit of putting out huge amounts of RFI-Hash (noise).
You may be having problems because of this.

D-9 is only rated for 1-Amp, it may already be smoked.
It needs to be rated substantially higher than the MEASURED full-load Amps of the Motor.

A Micro-Controller doesn't have enough Current Output to switch a large FET
with any respectable speed,
although the FET you are using is kinda wimpy.
It's probably running quite hot.

How are your Grounds physically arranged ?
Any Grounds using the Sheet-Metal Body of the Car ?
What kind of Temp-Sensor are you using ?
How is it Grounded ?
Is it a 1-wire, 2-wire, or 3-wire Sensor ?
Does the Temp-Sensor get its Ground from the Engine-Block ?
Are the Wires Shielded ?
Is your Micro-Controller mounted in a Metal-Box ?

Here are some ideas to make you think .................. (picture)View attachment 236957
i'm sorry i changed FR105 to FR305.
i use pwm duty maximum 80, so motor is notr over 3 Amp

i grounded car body, my temp sensor is NTC and it's 2wire it works good.
i grounded 2wire, 1->bettery, 2->car body

i have few questiong of your circuit sir

1. between motor+ and motor- there are so many cap.
but if i added cap in motor+ and motor- cap charging too much Amps
what is purpose that caps between motor+ and motor-?
is that for motor ripple?
because i had ripple because my pwm hz was too slow motor work with ripple
so i add 1000uF/50v cap between motor+, motor- and there was no ripple but cap is too much charging
how calculate cap value?

2. below IRL40B209, 10K R below, that diode. is that neccesery?

3. F=1(1.4 x R x C)
i'm using 16Mhz Crystal for MCU, and PWM is 8Khz
i want study that formula but i don't know what keyword,
is this formula about chosing cap value?

LowQCab

Joined Nov 6, 2012
517
1) These are small Ceramic Capacitors for suppressing RFI.
The FET sees these as a very tiny Load.
They are mostly for the protection of the FET,
the RFI Voltage Spikes can easily exceed twice the Power-Supply-Voltage.
( ask me how I know ....)
When using PWM do not put huge Caps across the Motor,
use an "Ultra-Fast" Diode, this will use the Inductance of the Motor Windings to
keep the Motor running during "off" periods.
The Faster this Diode is, the better it will work to smooth-out Motor operation.

2) There are 2-Diodes that protect the Input of the FET Driver,
These Diodes insure that the Input Voltage will not exceed the Power-Supply Voltage
that is supplying the FET Driver, by more than 0.7 Volts.
If that happens, it will smoke the FET Driver.
The PWM Circuitry will remain operational for up to ~1-Second after Power is removed,
making its output higher than the Power Supply of the FET Driver,
which will go to zero extremely fast when the Power Supply is removed.

3) The PWM Frequency of a Micro-Controller is determined by the Running-Program,
NOT the Operating Frequency of the Chip, or
the Crystal creating the "Clock" Signal.

The Formula is for selecting both the Capacitor, and the Resistor Values to
create a specific Oscillating-Frequency.
The Resistor-Value was kept at "10K" so that all of the Resistors would be the same Value.
The exact Frequency is not critical.
Generally it is desirable to have the frequency slightly above the audible range,
but when you start to go higher in Frequency,
the FET-Switching-Speed must be carefully calculated.

The PWM Circuitry is made from a single 8-Pin-DIP, Dual-Op-Amp,
which is operating as 2-Comparitors.
NOTE: Not all Op-Amps are suitable for use as a Comparitor.
This Circuit is vaguely similar to the operation of a 555 Timer-Chip.
"Zero" Speed is from zero Volts to 1/3 of VCC Voltage,
"Maximum" Speed is from 2/3 of VCC, to full VCC.
Varying Speeds, (varying Duty-Cycles), are from 1/3 to 2/3 of VCC,
( 3.3V to 6.6V with a Regulated VCC of 10-Volts ).

The Op-Amps are used instead of a "real" Comparitor because they have a
relatively low Output-Impedance which can supply ~20ma of Current,
this reduces the chance of outside-noise getting into the PWM Signal.
"Real" Comparitors usually have an extremely weak output,
designed for super-high-speed operation,
and not for driving a long wire in an extremely noisy environment.

The Heavy-Duty FET Driver insures that the FET will Switch "hard and fast",
this keeps the FET cool and increases Switching efficiency.
It also allows you to run a Big-Fat-High-Current-FET with lots of Gate-Capacitance,
with very small Switching-Losses.
Also, the FET Driver has a "Logic-Level" Input,
which your Micro-Controller's PWM-Output can easily drive.
Actually,
the FET Driver is stout enough to drive your small Motor directly,
without any FET at all, ( but I wouldn't risk it just to see ).

Your current set-up is very marginal in the FET Switching Department,
which might actually have some benefits,
because the slower Switching-Speeds will be less likely to cause unwanted oscillations.

This could cause the Chip to over-heat,
or cause internal Voltage-Spikes / Voltage-Sagging, or
smoke the whole Chip.
.
.
.

LowQCab

Joined Nov 6, 2012
517
i'm sorry i changed FR105 to FR305.
i use pwm duty maximum 80, so motor is notr over 3 Amp
The % Percentage of Duty-Cycle, or "Running" Current, of the Motor
does not affect the selection of the "Free-Wheeling-Diode".
The Diode needs to be sized based on the "Full-Load" or "Locked-Rotor" Amps
that the Motor Draws in a "worst-case" condition.
And then DOUBLE that Current-Rating for truly reliable operation.
.
.
.

Kim-JiHoon

Joined May 3, 2020
63
The % Percentage of Duty-Cycle, or "Running" Current, of the Motor
does not affect the selection of the "Free-Wheeling-Diode".
The Diode needs to be sized based on the "Full-Load" or "Locked-Rotor" Amps
that the Motor Draws in a "worst-case" condition.
And then DOUBLE that Current-Rating for truly reliable operation.
.
.
.
thank you so much sharing your knowledge!

first i tested this with my supply. and FR105 was smoked.

so i changed FR305 (this is my lagest Free Wheel diode).

I already order larger diode. (10A)

and i just tested different PWM frequency.

1. 5Khz - works good but motor got noise(strong high frequency sound)
2. 7Khz - works good but motor got noise(minimum high frequency sound)
3. 8Khz - works good but motor got noise(very low high frequency sound)
- but when i changed speed for test (such as 50% -> 80%, 80% ->10%)
motor goes full duty, even i programmed PWM max 80%.
4. 10Khz - works good and got zero noise.
- same as 8Khz

and 8,10Khz FET get too hot. (8Khz: 40 °C / 10Khz: 60 °C )

but fortunatly motor placed in box, so i don't need to worry about noise.

i don't need to use octocoupler to mosfet (mosfet mostly need gate voltage 12v~18v)

my motor is just 3A,so i found logic level fet and i connected directly to MCU.

until i got diode i'll test different resistors with FET.

i used 10~46R to open FET gate, i'll try 100R~250R.

and pull down resistor is 10K for this FET right?
(if pull down resistor value increase, discharging time will increase)

LowQCab

Joined Nov 6, 2012
517
Problems ..............

1) MAYBE, Electrical "Noise" getting into your Micro-Controller.
Usually caused by long Ground Wires, or high resistance in the Ground Wires,
or by running the Wires close to Electrical-Noise sources.
A Car is a VERY NOISY ELECTRICAL ENVIRONMENT.

2) Free-Wheeling-Diode,
The Diode you have on order will "probably" work just fine.
I would recommend an "ULTRA-FAST" Diode, such as a ........
FERD20S100STS , Digi-Key Part Number = 497-18626-ND , ~$0.90 cents. ( in your low-current application it will not require a Heat-Sink ) The Free-Wheeling-Diode should be Soldered directly to the Motor with very short Wires, or NO Wires. The Free-Wheeling-Diode can reduce RFI / Electrical-Noise which is produced at all times by the Arcing Motor Brushes. If you have long Wires between the Motor and the Free-Wheeling-Diode the Wires will act like an Antenna, which will broadcast RFI Noise into other systems. You also need small Ceramic Capacitors DIRECTLY SOLDERED TO THE MOTOR, for the same reasons. Multiple different sizes of Capacitors are desirable ........ 100nf, 10nf, 1nf, 100pf, each Capacitor will suppress a certain range of RFI Frequencies. One of these Capacitors "may" solve all major Noise problems, but determining which one will work is the "trick". 3) FET Gate Drive Current. You have inadequate Gate-Drive-Current. Your Micro-Controller is not capable of switching the Capacitance of the FET Gate Input at High-Frequencies. I would recommend this FET Gate-Driver ....... MIC4420ZT , Digi-Key Part Number , 576-2315-ND , ~$2.39
This Driver will require a ~10 to ~15-Volt, 1-Amp, Voltage Regulator to prevent
over-driving the FET Gate with 24-Volts.
A ~10-Volt Zener-Diode may be used to reduce the Voltage
to drop the maximum Voltage Supplied to the FET-Gate-Driver.
The FET-Gate-Driver requires a large Bypass-Capacitor, at least ~100uf.

4) The FET that you are using is rather "weak".
It has a 0.04 Ohm RDS rating, this can cause more Heat-Dissipation than is desirable.
I would recommend this FET ..........
TK100A08N1,S4X , Digi-Key Part Number , TK100A08N1S4X-ND , ~$3.93 It has twice the Voltage-rating of your current FET, which is extra insurance against Voltage-Spike damage, and it has an "RDS-ON" rating of 0.0039 Ohms, which means that, with a strong Gate-Drive-Current, it will dissipate only a very small amount of Heat. In your relatively low-current application it will "probably" not require a Heat-Sink, but a Heat-Sink is always good insurance in case something goes wrong. Recommended Heat-Sink ......... M47118B011000G , Digi-Key Part Number , 59-M47118B011000G-ND , ~$1.57,
The Gate Driver and the FET can both fit together on this single Heat-Sink.
No mounting-hardware is required, only Heat-Sink-Paste.

Some people will think that these recommendations are extreme "over-kill",
I simply look at it as extreme reliability, and stability, and
not having to smoke components 2 or 3 times until you get a circuit that actually works.

5) Gate Resistor ...........
I'm too lazy to go through all of the proper Modeling required to accurately
determine the "lowest-stable-value" Gate-Resistor,
so I will just make a "guestimation" and
say that a ~10-Ohm, 1-watt, Gate-Resistor will probably be just fine.
No Resistor to Ground is required for driving a Motor-Load,
but if you want to do things "by-the-book", and "properly",
then install a ~10K Resistor to Ground.

6) Micro-Controller ...........
If you don't need the special functions of a Micro-Controller, then don't use one.
A Micro-Controller "can be" very useful in many applications,
but if all you really need is a PWM, Thermostatic-Fan-Motor-Controller,
use Analogue / Linear Circuit design, instead of Digital control.
.
.
.

Last edited:

Kim-JiHoon

Joined May 3, 2020
63
Problems ..............

1) MAYBE, Electrical "Noise" getting into your Micro-Controller.
Usually caused by long Ground Wires, or high resistance in the Ground Wires,
or by running the Wires close to Electrical-Noise sources.
A Car is a VERY NOISY ELECTRICAL ENVIRONMENT.

2) Free-Wheeling-Diode,
The Diode you have on order will "probably" work just fine.
I would recommend an "ULTRA-FAST" Diode, such as a ........
FERD20S100STS , Digi-Key Part Number = 497-18626-ND , ~$0.90 cents. ( in your low-current application it will not require a Heat-Sink ) The Free-Wheeling-Diode should be Soldered directly to the Motor with very short Wires, or NO Wires. The Free-Wheeling-Diode can reduce RFI / Electrical-Noise which is produced at all times by the Arcing Motor Brushes. If you have long Wires between the Motor and the Free-Wheeling-Diode the Wires will act like an Antenna, which will broadcast RFI Noise into other systems. You also need small Ceramic Capacitors DIRECTLY SOLDERED TO THE MOTOR, for the same reasons. Multiple different sizes of Capacitors are desirable ........ 100nf, 10nf, 1nf, 100pf, each Capacitor will suppress a certain range of RFI Frequencies. One of these Capacitors "may" solve all major Noise problems, but determining which one will work is the "trick". 3) FET Gate Drive Current. You have inadequate Gate-Drive-Current. Your Micro-Controller is not capable of switching the Capacitance of the FET Gate Input at High-Frequencies. I would recommend this FET Gate-Driver ....... MIC4420ZT , Digi-Key Part Number , 576-2315-ND , ~$2.39
This Driver will require a ~10 to ~15-Volt, 1-Amp, Voltage Regulator to prevent
over-driving the FET Gate with 24-Volts.
A ~10-Volt Zener-Diode may be used to reduce the Voltage
to drop the maximum Voltage Supplied to the FET-Gate-Driver.
The FET-Gate-Driver requires a large Bypass-Capacitor, at least ~100uf.

4) The FET that you are using is rather "weak".
It has a 0.04 Ohm RDS rating, this can cause more Heat-Dissipation than is desirable.
I would recommend this FET ..........
TK100A08N1,S4X , Digi-Key Part Number , TK100A08N1S4X-ND , ~$3.93 It has twice the Voltage-rating of your current FET, which is extra insurance against Voltage-Spike damage, and it has an "RDS-ON" rating of 0.0039 Ohms, which means that, with a strong Gate-Drive-Current, it will dissipate only a very small amount of Heat. In your relatively low-current application it will "probably" not require a Heat-Sink, but a Heat-Sink is always good insurance in case something goes wrong. Recommended Heat-Sink ......... M47118B011000G , Digi-Key Part Number , 59-M47118B011000G-ND , ~$1.57,
The Gate Driver and the FET can both fit together on this single Heat-Sink.
No mounting-hardware is required, only Heat-Sink-Paste.

Some people will think that these recommendations are extreme "over-kill",
I simply look at it as extreme reliability, and stability, and
not having to smoke components 2 or 3 times until you get a circuit that actually works.

5) Gate Resistor ...........
I'm too lazy to go through all of the proper Modeling required to accurately
determine the "lowest-stable-value" Gate-Resistor,
so I will just make a "guestimation" and
say that a ~10-Ohm, 1-watt, Gate-Resistor will probably be just fine.
No Resistor to Ground is required for driving a Motor-Load,
but if you want to do things "by-the-book", and "properly",
then install a ~10K Resistor to Ground.

6) Micro-Controller ...........
If you don't need the special functions of a Micro-Controller, then don't use one.
A Micro-Controller "can be" very useful in many applications,
but if all you really need is a PWM, Thermostatic-Fan-Motor-Controller,
use Analogue / Linear Circuit design, instead of Digital control.
.
.
.View attachment 237051
i'm sorry i'm late.

i just ordered ICs which you recommanded. it take two or three weeks. thank you

and i tested some circuit.

i tested R18 10R, R19 4.7K ~ 10K

but i don't feel different between R18 10R, 47R

i thought if i use 10R, PWM signal can't fully open FET gate. and some times fet gets fully open even PWM signal 0

so now i use R18 46R and R19 10K, it works good. and FET no heat. (as you said)

i moved my FR diode to DC MOTOR directly.
(before, diode was in controller pcb and wiring 1M to motor)

MCU is 16Mhz (16000000 / 6 / 250= 8,000hz)
PWM hz is 8Khz

so far i tested so many ways. there was no problem!!

but i found little issue here

C2 is little warm about 30 °C ~ 32 °C

MOTOR+ is connected +24V in PCB.

i think C2 is quite large capacitor, and cause of ripple on my input vcc.

should i change C2 1000uF -> 470uF/50v + 470uF/50v ?

i worried C2 blow out.

wayneh

Joined Sep 9, 2010
17,152
i think C2 is quite large capacitor, and cause of ripple on my input vcc.
Is the 7805 warm? All the current going thru the capacitor is also going thru the regulator and the load, which should be fairly small? I'm a little surprised you're seeing warming there.
should i change C2 1000uF -> 470uF/50v + 470uF/50v ?

i worried C2 blow out.
There are several potential solutions. My preference would be to find a low ESR cap (lower heat production for a given ripple current) with a good temperature rating, eg. 105°C. Larger physical size can help a bit also, and a shape with greater L/W ratio.

LowQCab

Joined Nov 6, 2012
517
C-2 will never "cause" ripple.
C-2 is completely unnecessary, remove it from the Circuit.
IF D-3 is shorted, (damaged), it could cause overheating of C-2.
C-2 should never be hot, or even warm, unless there is a problem with D-3.

With an SMD Board, sometimes it is very difficult to determine which part is getting hot.
Heat can transfer all across the Board, and from top to bottom.
Heat is one of the biggest problems with SMD Boards.

C-9 is much larger than it needs to be. Use a 2.2uf Tantalum, plus a 100nf Ceramic.
Look at the LM-7805 Data-Sheet for Input and Output Capacitor recommendations.
Large Bulk-Storage-Capacitors are not necessary for a Micro-Controller Power-Supply.
High-Frequency Capacitors are important to reduce Oscillations.
Electrolytic-Caps are slow, they are for "Bulk-Storage",
Tantalum-Caps are much faster, and
Ceramic-Caps are High-Frequency, (very fast),
(but not all Ceramics are equal, there are many different types).

C-12 can be a 2.2uf Tantalum, but the 100uf Electrolytic-Cap is probably not a problem.

R-19 only has one purpose .........
It is there to insure that the Gate is not "Floating" when the Power is OFF.
This will insure that the Gate is at "zero" Volts when the Circuit is First Powered-Up.
It also protects the Gate from "Static-Electricity" when the Circuit is not powered.
It is NOT a "Pull-Down" Resistor.
The Micro-Controller provides both the "Source", and "Sink", Voltage/Current for the Gate.
This resistor can create an addition Load for the PWM Output to Drive,
If you make this Resistor too low, you will reduce the Gate-Drive-Voltage to the FET.
It should not be smaller than ~10K, and 47K would be just right.

From the Data-Sheet ..................
""" Each output buffer has symmetrical drive characteristics with
both high sink and source capability.
The pin driver is strong enough to drive LED displays directly """"""
This is the only information I could find in the Data-Sheet that gives any indication
of the PWM-Output-Current capability.
In my estimation ..........
strong enough to drive LED displays directly
means that the Outputs can Source or Sink approximately ~10ma., (if you are lucky),
the output is probably down around ~5ma. in reality.

These Pictures are from a Circuit-Simulation Software-Program,
so the exact Wave-Form representations are not 100% accurate.
But the program is very useful in finding Circuit Problems,
like the problems that you are having right now.

R-18 is only to reduce Oscillations caused by the Gate-Capacitance of the FET when there
may be TO MUCH Current available for Gate Drive.

But, the problem is ........
Your PWM Output HAS INTERNAL RESISTANCE that is way too HIGH.
It has a very limited capacity to drive a Heavy-Capacitive-Load like a Large FET Gate.

This problem makes R-18 almost irrelevant under the present situation.
A 10-Ohm Resistor will do nothing in this position
when the PWM-Output "may" have an Output Impedance as high as ~1K-Ohms.

The only thing R-18 is doing, in your Circuit,
is slightly reducing the Drive-Voltage to the FET Gate,
which is already too low of a Voltage, (and Current), for efficient operation.
.
.
.